Design and analysis of a parallel file system for distributed shared memory systems

File accesses are usually sequentially performed in existing Distributed Sh ared Memory (DSM) systems. These sequential file accesses will result in th e accumulation of input data at the node that handles file operations, gene rating a bottleneck at that node and a large amount of network traffic to m ove the input data to other nodes for execution. Although the file access t ime is often neglected in the performance evaluation of DSM systems, it is not ignored by programmers in real life. In this paper, we will describe th e design and analysis of a parallel file system for DSM systems. File acces ses are carried out in parallel and a modified file access mechanism is pro vided to reduce network traffic. Our analysis shows that the overall perfor mance of some I/O-intensive DSM applications such as Successive Over Relaxa tion (SOR) can be significantly enhanced with our design in the best case. To approach this best case, we have proposed two implementations. Both impl ementations employ variable-distribution schemes, which distribute file blo cks on the disks according to the access pattern of the application. (C) 19 99 Elsevier Science B.V. All rights reserved.